More experienced teachers are likely to benefit students. A developing body of research finds teachers with more years of teaching experience increase student growth on achievement tests more than novice teachers (Kini & Podolsky, 2016; Ladd & Sorensen, 2017). New teachers can educate, mere experience does not produce excellence, and teaching is just one of many factors influencing school outcomes, but on average, researchers frequently find a positive relationship between years of experience and student outcomes. Numerous studies have also concluded that teacher experience, and teacher “quality” more broadly, is racially distributed across the nation and specifically in California (Darling-Hammond, 2004; Goldhaber et al., 2015; Knight, 2019; Shields et al., 1999). Schools with high proportions of Black or Latino students have a disproportionate share of inexperienced teachers.

Several policies and initiatives have attempted to address the inequitable distribution of experience, including the Every Student Succeeds Act (2015) and other influential federal policies in recent years (Knight, 2019). State-level policies and analyses in California have addressed the racial distribution of teacher experience, including the State’s release of three annual reports in 2015, 2016, and 2017 titled California’s State Plan to Ensure Equitable Access to Excellent Educators. Public pressure, including a high-profile court case challenging California’s staffing decisions, Vergara v. California, may have influenced the distribution of teacher experience. Given the importance of teacher experience and possible changes to the racial distribution of experience in recent years, this policy brief examines the distribution of teacher experience across segregated schools in California.

The analysis used 7 years of school-level student race and teacher experience data in California public schools. Teacher experience was defined in several ways, including designating teachers as novice at three thresholds (1, 2 or 3 or fewer years teaching experience). The mean and median years of teacher experience in each school was also examined. Segregation was measured from multiple standpoints, where the racial composition of schools was used to group schools in five ways. This included designating schools as majority non-white and 90% or more under-represented minorities. The Appendix contains more details on the data and measures used.

Several findings are relevant for policymakers and school stakeholders:



Figure 1: Percent of Novice Teachers, By Student Segregation (2012-13 to 2018-19)

One or Fewer Years Experience

Two or Fewer Years Experience

Three or Fewer Years Experience

## {.unlisted .unnumbered}

Data

Teaching experience varied with student segregation (Figure 1). In each of the 7 years analyzed, majority white schools had fewer novice teachers than schools with higher proportions of racially marginalized students. In particular, schools with a student composition that was 90% or greater under-represented minorities had a higher proportion of novice teachers than schools that were majority white or majority white plus Asian. That teacher experience gap doubled from approximately 1.5 percentage points to 3 percentage points between 2012-13 and 2018-19 (for 2 or fewer years experience).

The proportion of teachers with 1, 2 and 3 years of experience or fewer was analyzed (see the separate panels in Figure 1). The results for all three definitions of novice were similar: majority white and majority white plus Asian schools had lower proportions of novice teachers than majority non-white, 90-100% non-white, and 90-100% underrepresented minority schools. In alternative measurements, the gap began to widen in approx. 2014-15. When defining novice as the proportion of teachers with 3 or fewer years of experience, the experience gap between majority white and majority non-white was approx. 4 percentage points in 2018-19.



Table 1: California Schools with a Majority Novice Faculty (2018-19)

There were 119 schools in 2018-19 where the majority of teachers had 2 or fewer years of teaching experience (Table 1). These schools varied in terms of student segregation, with some schools reporting no white students to one school that was 72.8% white. Approximately 79.8% of schools with a majority novice faculty were disproportionately non-white (higher than the white enrollment in the state of approx. 24%). In other words, there was a high concentration of inexperienced teachers in schools with a high concentration of students of color.



Robustness Checks: Alternative Measures of “Novice”

The finding that schools with the highest proportions of racially marginalized students have the least experienced teachers could be an artifact of how teacher experience was measured. Several robustness checks, however, suggest this is not the case. The educational returns to additional years of experience are likely greatest in the first few years of teaching (Kini & Podolsky, 2016), so teachers with a “few years under their belts” are more likely to promote academic growth than new teachers. The primary analysis above used the proportion of teachers with 2 or fewer years of experience. However, because some recent reviews of research find that additional teaching experience is associated with student gains into teachers’ second and even third decade (e.g., Kini & Podolsky, 2016), just looking at the proportion of new teachers may not usefully describe the extent of teacher experience gaps. Several alternative measures were created to see if the central finding was robust to alternative measures of teaching experience. These alternative measures included the proportion of teachers with 1 and 3 years of experience (see above) and the average of teacher experience in schools (see below). Regardless of the type of measurement, similar teacher experience gaps were identified across racially-identifiable schools.



Figure 2: Mean Years of Teaching Experience and Student Racial Composition

2012-13

2013-14

2014-15

2015-16

2016-17

2017-18

2018-19

## {.unlisted .unnumbered}

The scatterplots in Figure 2 depict the relationships between mean years of teaching and the percent of the five focal racial groups. Each dot represents a California school in 2018-19. Average years of teaching experience was negatively correlated with Black, under-represented minority, and Hispanic enrollments and positively correlated with white and Asian enrollments. The relationship was strongest with black enrollment and rather weak for the other racialized groups (see the blue best fit lines). The most positive relationships with mean years of teaching experience was the proportion of white and Asian students. Pooling the data from 2012-13 to 2018-19 created similar results as did each year run separately. The same analysis conducted with median years of experience, instead of mean years of experience, produced similar trends, with slightly larger gaps across racial groups.



Figure 3: Mean Teacher Experience in the Most Racially Concentrated Schools (2018-19)

Note: Figure includes the top quintile or 20% of schools (approx. 1,646) in each racial category.

A comparison of schools with the greatest of various racial groupings is presented in Figure 3, which permits an additional way to explore differences in teacher experience across segregated schools. Schools were designated in the highest quintile if they were in the top one-fifth of schools by racial concentration. For example, schools with enrollment greater than 44.1% white students were the top 20% whitest schools. The mean years of experience for those schools was 14.6. This was 1.4 years more teacher experience than in the blackest schools, where mean years of teacher experience was 13.2.

The gap was fairly consistent across the years studied, ranging from a white advantage of approx. 1.3 years in 2013-14 to a white advantage of 1.5 years in 2017-18. Over the time period studied, 2012-13 to 2018-19, the black-white teacher experience gap did not improve. In each year studied, schools with the highest concentrations of white, white plus Asian, and Asian students employed more experienced teachers than schools with the highest concentrations of Hispanic, underrepresented minority, and black students.

Figure 4: Teacher Experience By Student Segregation (2018-19)

Majority white schools had slightly more than 1 year of additional mean experience than schools that enrolled more than 90% under-represented minorities (Figure 4).

Illustrative Example of a School that Became Whiter Each Year

Note: Focal schools with anomalous data were removed, including schools that changed radically in terms of racial composition or mean teacher experience from one year to the next.

The figure above shows an example of a school that increased in proportion white each year and increased in mean teacher experience in 4 out of 5 years. This kind of association, if systematic, would be evidence that segregation may significantly pressure and limit teacher experience. There are, of course, other schools that show the opposite association. But the focus, from a policy perspective, should likely be on prevailing trends. The analysis below investigates whether there are systematic trends that are consistent with this illustrative school.

Correlation between White Change in 7 years and mean years experience over the same 7 years

Was an increase in white enrollment associated with an increase in teacher experience over the 7 years studied?

The correlation between the white student proportion enrollment change from 2012 to 2018 and the mean years of (mean)teacher experience change over the same time period was 0.06. This suggests that in schools that became whiter, teacher experience also increased.

Beyond simple correlation: Connecting Segregation to Teacher Experience

Some readers may believe that with sufficient policy reform, the distribution of teacher experience can be improved in racially segregated schools. A different set may believe that while reforms can result in improvements for segregated schools, a necessary pre-requisite for a substantial and durable equitable distribution of teacher experience, desegregation is required. This latter belief has substantial correlational evidence on its side: the analysis above and others find that segregated schools consistently have less experienced teachers. There is other evidence (CITE) that segregated schools are indeed different in a number of ways that may make it more difficult for them to employ experienced teachers. This section investigates whether there is a link between segregated schools and teacher inexperience, beyond mere cross-sectional correlation.

The following causal question is difficult to answer, especially without a randomized control trial or significant qualitative research: Does school segregation cause lower teacher experience? However, the question sharpens what would be good to know. We can begin to address this causal question by trying to hold many possible influences on the distribution of teacher experience equal and see if the positive correlation between teacher experience and segregation holds. To do this, we control for observed characteristics of schools, but perhaps more importantly, we can control for unobserved characteristics by employing a first-differences modeling approach.

Perhaps more importantly than including controls of observed variables, our first-differences approach controls for unobserved time-invariant factors, which reasonably may include neighborhood characteristics, labor market conditions, urbanicity, union strength, and many other factors which may influence the distribution of teacher experience. For example, districts’ transfer policies may influence where teachers work, but because we only model experience changes within schools, this district influence is “held constant.” School architecture is another example of how our first-differences approach controls for unobserved time-invariant factors. Old buildings in disrepair may make teachers want to leave and we may find levels of disrepair connected to racial composition. We do not have data on school architecture, but as long as schools do not systematically change architecture in ways that teacher experience and racial composition, then school architecture will not bias the results. In sum, omitted variable bias would only occur if unobserved factors systematically vary with mean teacher experience and racial composition.

We modeled longitudinal data with observations within the same schools across multiple years. The longitudinal data allowed leveraging temporal ordering, as well as use a within-school estimator, to reduce the likelihood of omitted variable bias, and present a rigorous estimation of the relationship between racial composition and teacher experience. Using a first-difference analytic method, we regressed year-over-year changes in mean teacher experience (\(\\\Delta\overline{Experience}_{it}\)) by changes in school racial composition. A first-difference approach is similar to the commonly used fixed-effects model and shares its primary attraction: controlling for unobserved, time-invariant school characteristics which could contribute to omitted variable bias (Cameron and Trivedi 2005). Equation (1) describes the first-difference model,

\[\begin{equation} \begin{aligned} \Delta\overline{Experience}_{it} = \beta_{0} \\ &+ \beta_{1} \Delta\mathrm{\%Black}_{it} \\ &+ \beta_{2} \Delta\mathrm{\%Hispanic}_{it} \\ &+ \beta_{3} \Delta\mathrm{\%Asian}_{it} \\ &+ \beta_{4} \Delta\mathrm{\%MultiRacialandOther}_{it} \\ &+ \beta_{5} \Delta\mathrm{Enrollment}_{it} \\ &+ \mathrm{\lambda}_{i} \\ &+ \epsilon_{it} \end{aligned} \tag{1} \end{equation}\]

where \(\Delta\mathrm{\%Black}_{it}\) represents the change in proportion of Black students, \(\Delta\mathrm{\%Hispanic}_{it}\) represents the change in proportion of Hispanic students, \(\Delta\mathrm{\%Asian}_{it}\) represents the change in proportion of Asian students, \(\Delta\mathrm{\%MultiRacialandOther}_{it}\) represents the change in proportion of Multi-Racial and Other students. Percent white (omitted) is the reference category.

The main coefficients of interest are the average changes in the proportion of each racial group associated with a 1 year change in mean teacher experience (compared to White students). If there were no time-varying omitted variables which are correlated with both change in racial composition and change in mean teacher experience, these would be closer to unbiased estimates (Liker, Augestyniak, and Duncan 1985).

Year fixed effects, \(\mathrm{\lambda}_{i}\), controlled for unmeasured year-over-year effects, invariant across schools. \(\epsilon_{it}\) was the error term. Standard errors for Equation (I) were clustered at the school level to account for the dependence of repeated school observations over time.

[subsequent models could include these:] Because omitted variable bias was an important threat to identifying a causal relationship, we included several control variables which were plausibly related to teacher experience and racial composition: student’s FRL, ELL, etc

Looking closer at causation: Results

The regression results below have three models. The first is the result from equation 1. The others are variations. All show that, on average , schools with an increase in the shares of Black or Hispanic students, were negatively associated with mean teacher experience, holding much else equal. For example, the model of Eq. 1 suggests that as the percent Black increased 10 percentage points, the mean teacher experience decreased by 0.13 years. Depending on the model specification the results meet various levels of statistical significance.

plm1 plm2 plm3
stu_Hispanic_annual_diff −0.005 −0.012*** −0.010***
(0.004) (0.003) (0.003)
stu_AA_annual_diff −0.013+ −0.020** −0.017**
(0.007) (0.006) (0.005)
stu_Asian_annual_diff −0.008 −0.016** −0.014**
(0.007) (0.006) (0.005)
stu_MultiRacialandOther_annual_diff 0.007
(0.005)
Total_Enrollment 0.000*** 0.000*** 0.000*
(0.000) (0.000) (0.000)
stu_White_annual_diff −0.007 0.000
(0.005) (0.004)
(Intercept) 0.005
(0.011)
Num.Obs. 47039 47039 47039
R2 0.001 0.001 0.001
R2 Adj. −0.220 −0.220 0.001
AIC 150021.9 150021.9 158674.3
BIC 150030.7 150030.7 158683.0
RMSE 1.19 1.19 1.31
+ p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001

Limitations

Schools with equal mean years of experience (or percent with < X years experience) are not necessarily equal in important types of experience. Imagine two schools, both with 10 years of mean teacher experience. One school has had the same teachers from the prior year, and the other has 10 teachers new to that school, which is a form of churn between schools. The school with greater teacher turnover has higher instability that is may exact costs. Disruption and lack of continuity in the relationships teachers build with their particular school communities, students, staff, norms and school policies is associated with an array of problems (Simon & Johnson, 2015). Additionally, schools with greater teacher turnover incur higher financial costs as they must find new teachers more often. Furthermore, churn within schools and across grades or subjects may be important, related to between-school churn, and associated with student demographics (Atteberry et al., 2017; Huang & Moon, 2009).

Schools with small enrollments and small numbers of teachers can make measures of novice teachers and measures of racial segregation vary somewhat noisily.

Discussion

This analysis identified persistent gaps in teacher experience across segregated schools. Schools with higher proportions of marginalized racial groups had a higher proportion of novice teachers and the trend worsened over time. The trends identified may constitute opportunity gaps that cause achievement gaps. The size of these gaps, which may appear small to small readers, may become practically significant when considering that the effects of teacher (in)experience can accumulate over 12 or more years of schooling.

There are multiple possible mechanisms through which teacher experience may improve schooling. On-the-job experience may improve teachers’ pedagogical knowledge, learning goals, and classroom practices. Faculties with greater experience may also have positive school-wide effects through improved support networks and mentoring among teachers. Regardless of the specific mechanisms, research consistently finds that more experienced teachers are associated with improved student outcomes.

The inequitable distribution of teacher experience in California mirrors national trends. National research finds that the most inexperienced teachers are clustered in schools with the most marginalized students, with racialized minority students more likely to employ “green” teachers (Cardichon et al., 2020; Knight, 2019). Opportunity gaps across race exist in several types of school resources (Carter & Welner, 2013). One interpretation of equitable distribution of school resources suggests that the most marginalized students would receive the most resources to promote equal opportunity. Higher concentrations of the least experienced teachers in racially marginalized schools, which our analysis finds is the case in California, runs counter to that definition of equal opportunity.

With a growing body of research concluding that teacher experience improves school success, teacher experience gaps across segregated schools constitutes part of an ongoing opportunity gap in California. Efforts to address the distribution of teacher experience have been made in recent decades, yet the trends observed in this analysis suggest different approaches are required to improve opportunities to learn for racially marginalized children in California.

Teacher experience gaps may be decreased by reforming several types of policies. Changing teacher tenure, seniority preference, transfer, and dismissal practices may help slightly (Goldhaber et al. (2016)), yet at least two empirical studies suggest other directions may be required (Knight (2019); Koski & Horng (2007)). One study (Knight (2019)) finds additional funding is associated with lower teacher experience gaps across high- and low-poverty schools, although the effect across racially segregated schools is not clear. Another study finds “no persuasive evidence that the seniority preference rules… exacerbate the negative relationship between higher minority schools and uncredentialed and low-experience teachers” (Koski & Horng (2007), p.262).

Given the rather inconclusive evidence on the significance of factors related to human capital management and collective bargaining as causes of teacher experience gaps, other types of policy changes may need to be tried (and studied). One related approach involves offsetting the negative effects of inexperienced teachers through other school reforms such as lowering class sizes in high minority schools. Other possibilities include: changing the distribution of school funding; macroeconomic reforms that affect the labor market competition among high minority schools and other sectors of the economy; and increasing the prestige, economic reward, and well-being that comes with teaching in schools with high proportions of underrepresented children of color.

Regardless of which reforms may best improve the distribution of teacher experience, the best reforms are likely to pay particular attention to racial segregation. The evidence presented in this report adds to a significant body of research concluding that equal educational opportunity is thwarted by school segregation. Addressing racial segregation and the distribution of teacher experience simultaneously may produce the most substantial benefits.


References


Atteberry, A., Loeb, S., & Wyckoff, J. (2017). Teacher churning: Reassignment rates and implications for student achievement. Educational Evaluation and Policy Analysis, 39(1), 3–30.
Cardichon, J., Darling-Hammond, L., Yang, M., Scott, C., Shields, P. M., & Burns, D. (2020). Inequitable opportunity to learn: Student access to certified and experienced teachers. Learning Policy Institute.
Carter, P. L., & Welner, K. G. (2013). Closing the opportunity gap: What america must do to give every child an even chance. Oxford University Press.
Darling-Hammond, L. (2004). Inequality and the right to learn: Access to qualified teachers in california’s public schools. Teachers College Record, 106(10), 1936–1966.
Goldhaber, D., Lavery, L., & Theobald, R. (2015). Uneven playing field? Assessing the teacher quality gap between advantaged and disadvantaged students. Educational Researcher, 44(5), 293–307.
Goldhaber, D., Lavery, L., & Theobald, R. (2016). Inconvenient truth? Do collective bargaining agreements help explain the mobility of teachers within school districts? Journal of Policy Analysis and Management, 35(4), 848–880.
Huang, F. L., & Moon, T. R. (2009). Is experience the best teacher? A multilevel analysis of teacher characteristics and student achievement in low performing schools. Educational Assessment, Evaluation and Accountability, 21(3), 209–234.
Kini, T., & Podolsky, A. (2016). Does teaching experience increase teacher effectiveness? A review of the research. Learning Policy Institute.
Knight, D. S. (2019). Are school districts allocating resources equitably? The every student succeeds act, teacher experience gaps, and equitable resource allocation. Educational Policy, 33(4), 615–649.
Koski, W. S., & Horng, E. L. (2007). Facilitating the teacher quality gap? Collective bargaining agreements, teacher hiring and transfer rules, and teacher assignment among schools in california. Education Finance and Policy, 2(3), 262–300.
Ladd, H. F., & Sorensen, L. C. (2017). Returns to teacher experience: Student achievement and motivation in middle school. Education Finance and Policy, 12(2), 241–279.
Shields, P. M., Esch, C. E., Humphrey, D. C., Young, V. M., Gaston, M., & Hunt, H. (1999). Teaching and california’s future. The status of the teaching profession: Research findings and policy recommendations.
Simon, N., & Johnson, S. M. (2015). Teacher turnover in high-poverty schools: What we know and can do. Teachers College Record, 117(3), 1–36.


Appendix: Data and Measures

Data analyzed in this report came from publicly available student and staff files provided online by the California Department of Education (CDE). The analysis covers all school years made available by CDE with teacher data: 2012-13 to 2018-19.

The following rules were applied to prepare the data for analysis:

The final analytic data set included an average of 8,118 schools per year.

Measure of Experience

One measure of teacher experience used was the proportion in each school of teachers with 2 or fewer years of teaching experience. That is, the proportion of schools’ teachers in their first or second year of teaching. Alternative specifications include teachers with 1 and 3 years or fewer experience. Experience, as measured by the CDE, includes prior teaching experience in any school, including outside of California. Experience substitute teaching or classified staff service is not included, according to definitions provided by CDE. A teacher in their first year of employment is categorized by CDE as having 1 year of experience.

Measures of Student Segregation

To examine teacher experience gaps across different types of segregated schools, schools were grouped into one or more of the following categories: majority white, majority white and Asian, majority non-white, 90-100% non-white, and 90-100% Under-Represented Minority (URM). “Majority white and Asian” included Filipino. Non-white was defined as a sum of all the racial groups provided by CDE other than white (i.e., Hispanic, Black, Asian, American Indian, Filipino, Pacific Islander, and Two or More races). Under-Represented Minority was defined as the sum of students classified as Hispanic, Black, American Indian, Pacific Islander, or Two or More races.




The Civil Rights Project